COVID-19 Analysis across countries and weeks

In this study, the focus is on the country cases. This analysis examines at the case growth, case proportion and weekly growth.

This kerenel will be updated frequently to keep it up to date

Library and dataset imports

In [1]:
#Importing the libraries
import pandas as pd
import numpy as np

import matplotlib.pyplot as plt
import plotly.express as px
from plotly.subplots import make_subplots
import plotly.graph_objects as go

The data used for this study is taken from the Johns Hopkins CSSE data repository. The following files were used for the analysis:

  1. time_series_covid19_confirmed_global.csv-> https://rb.gy/uktxf3
  2. time_series_covid19_deaths_global.csv -> https://rb.gy/qnjgsj
  3. time_series_covid19_recovered_global.csv-> https://rb.gy/dxfjfl
In [2]:
#Importing the datasets
url='https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv'
confirmed=pd.read_csv(url,error_bad_lines=False)
death=pd.read_csv('https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_global.csv',
                 error_bad_lines=False)
recovered=pd.read_csv('https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_recovered_global.csv',
                 error_bad_lines=False)

Case proportions of countries

The function country_with_cases will return a dataframe with just the country name and the number of cases. This function also adds up all the province value into one to make the overall data just in terms of the Country/Region. The main reason this is done is because the Province/State contains a lot of null values

In [3]:
def country_with_cases(dataset):
    is_province=dataset.loc[dataset['Province/State'].isna()==False]
    is_province=is_province['Country/Region'].unique()
    df=dataset.copy()
    Country=df['Country/Region'].values
    temp=df.drop(columns=['Province/State','Country/Region','Lat','Long'])
    values=temp.values
    cases=[]
    for i in range(0,len(values)):
        cases.append(values[i][values.shape[1]-1])
    new_df=pd.concat([pd.DataFrame(Country),pd.DataFrame(cases)],axis=1)
    new_df.columns=['Country','Cases']
    index=[]
    is_province_sums=[]
    for i in is_province:
        temp=new_df.loc[new_df['Country']==i]
        index.append(temp.index)
        s=np.sum(temp['Cases'])
        is_province_sums.append(s)
    for i in index:
        new_df.drop(i,axis=0,inplace=True)
    countries_with_province=pd.concat([pd.DataFrame(is_province),pd.DataFrame(is_province_sums)],axis=1)
    countries_with_province.columns=['Country','Cases']
    All_country_cases=pd.concat([new_df,countries_with_province],axis=0)
    All_country_cases.reset_index(inplace=True)
    return All_country_cases
In [4]:
confirmed_tree=country_with_cases(confirmed)
In [5]:
def tree_map(df,color_scale,title):
    #The df is the one which only has country and cases
    fig = px.treemap(df,path=['Country'], values='Cases',color='Cases',
                     color_continuous_scale=color_scale,
                     title=title)
    fig.show()
In [6]:
tree_map(confirmed_tree,'amp','Confirmed cases across countries')
In [7]:
recovered_tree=country_with_cases(recovered)
tree_map(recovered_tree,'Greens','Recovery across countries')
In [8]:
death_tree=country_with_cases(death)
tree_map(death_tree,'Reds','Deaths across countries')

Visualising the Country growth

In [9]:
def top_10(df):# This function will only work 
    df_descending=df.sort_values(by='Cases', ascending=False)
    df_descending=df_descending.reset_index()
    top=df_descending.iloc[:10 :]
    return top['Country'].values

The rate function converts the structire of the dataframe in general. In the dataset the dates are the column making it not so convenient for visualising the data. This function transforms all the dates into one column and this makes plotting very much easier

In [10]:
def rate(df):   
    is_province=df.loc[confirmed['Province/State'].isna()==False]
    is_province=is_province['Country/Region'].unique()
    copy=df.copy()
    final=[]
    index=[]
    for i in is_province:    
        temp=copy.loc[copy['Country/Region']==i]
        index=copy.loc[copy['Country/Region']==i].index
        temp=temp.sum(axis=0)
        final.append(temp)
        copy.drop(index,inplace=True)
    new_df=pd.DataFrame(final)
    new_df['Country/Region']=is_province
    total=pd.concat([copy,new_df],axis=0)
    total.reset_index(inplace=True)
    total.drop(columns=['Province/State'],inplace=True)
    t=pd.melt(total,id_vars=['Country/Region','index','Lat','Long'],var_name="Date", value_name="Value")
    return t
In [11]:
c=rate(confirmed)
fig = px.line(c, x="Date", y="Value", color="Country/Region",
              title='Confirmed cases across countries')
fig.update_layout(showlegend=False)
fig.show()
In [12]:
r=rate(recovered)
fig = px.line(r, x="Date", y="Value", color="Country/Region",
              title='Confirmed cases across countries')
fig.update_layout(showlegend=False)

fig.show()
In [13]:
d=rate(death)
fig = px.line(d, x="Date", y="Value", color="Country/Region",
              title='Deaths across countries')
fig.update_layout(showlegend=False)

fig.show()

Analysing the top 10 countries

We will analyse 10 countries with the highest confirmed case

In [14]:
top_ten=top_10(confirmed_tree)
In [15]:
def stacked_line_subplots(confirmed,death,recovered,countries):
    subplot_title=[]
    for i in countries:
        subplot_title.append('Cases in {}'.format(i))
    subplot_title=tuple(subplot_title)
    fig = make_subplots(rows=len(countries), cols=1,subplot_titles=subplot_title)
    #countries=['India','US','Yemen','Angola']
    dates=confirmed.columns
    dates=np.delete(dates,[0,2,3])
    dfs=[confirmed,death,recovered]
    row=1
    for i in range(len(countries)):
        value=[]
        for j in dfs:
            temp=j.loc[j['Country/Region']==countries[i]].values
            temp=np.delete(temp,[0,1,2,3])
            value.append(temp)
        if(i==0):
            fig.append_trace(go.Scatter(x=dates,y=value[1],mode='lines',name='Death',
                         line_color='red',stackgroup='covid',legendgroup="group1"),row=row,col=1)
            fig.append_trace(go.Scatter(x=dates,y=value[2],mode='lines',name='Recovered',
                         line_color='green',stackgroup='covid',legendgroup="group2"),row=row,col=1)
            fig.append_trace(go.Scatter(x=dates,y=value[0],mode='lines',name='Confirmed',
                         line_color='blue',stackgroup='covid',legendgroup="group3"),row=row,col=1)
        else:
            fig.append_trace(go.Scatter(x=dates,y=value[1],mode='lines',name='Death',
                         line_color='red',stackgroup='covid', showlegend=False,
                                        legendgroup="group1"),row=row,col=1)
            fig.append_trace(go.Scatter(x=dates,y=value[2],mode='lines',name='Recovered',
                         line_color='green',stackgroup='covid', showlegend=False,
                                       legendgroup="group2"),row=row,col=1)
            fig.append_trace(go.Scatter(x=dates,y=value[0],mode='lines',name='Confirmed',
                         line_color='blue',stackgroup='covid', showlegend=False,
                                       legendgroup="group3"),row=row,col=1)
            

        row+=1
        
        fig.update_layout(height=2000, width=800,
                  title_text="Cases across the top 10 countries")
            
    fig.show()

This stacked line graph in a way shows the ratio of the cases in a country. The blue area in way indicates the active cases(when both red and green areas are present), green indicates recovered and red indicates the deaths

In [16]:
stacked_line_subplots(confirmed,death,recovered,top_ten)

Visuals on the World Map

In [17]:
def world_map(df,title,color):
    dates=df.columns
    dates=np.delete(dates,[0,2,3])
    country=df['Country/Region']
    lat=df['Lat']
    long=df['Long']
    transformed=pd.melt(df,id_vars=['Province/State','Country/Region','Lat','Long'],
                   var_name="Date", value_name="Value")
    fig = px.scatter_geo(transformed, lat="Lat",lon="Long", color="Value",
                     hover_name="Country/Region", size=transformed["Value"],
                     animation_frame="Date",
                     projection="natural earth",title=title,
                     color_continuous_scale=color)
    fig.show()
In [18]:
world_map(confirmed,'Confirmed cases with time series','Jet')
In [19]:
world_map(recovered,'Recovered cases with time series','YlGn')
In [20]:
world_map(death,'Death Cases with time series','Burg')

Weekly analysis on case growth

The function weekly_trend() mainly creates a new column named week. This column indicates the week number of a particular day. For this, the dates have been converted to days since the first reported day. The function will return the a dataframe of one particular week

In [21]:
from datetime import date
def weekly_trend(df,number):
    c=rate(df)
    date_values=c['Date'].values
    first_date=date_values[0].split('/')
    first=list(map(int, first_date))
    for i in range(len(date_values)):
        random=date_values[i].split('/')
        random=list(map(int, random))
        delta=date(2020,random[0],random[1])-date(2020,first[0],first[1])
        date_values[i]=delta.days
    c['Date']=date_values
    c['Weeks']=c['Date']//7
    confirmed_week=c.loc[c['Weeks']==number]
    confirmed_week=confirmed_week.sort_values(by='Value',ascending=False)    
    return confirmed_week

The function top_3_trends returns the top 3 countries. These are the countries which have the highest growth in the week

In [22]:
def top_3_trends(week,week_number):
    last_day=week.loc[week['Date']==((week_number*7)+6)]
    first_day=week.loc[week['Date']==(week_number*7)]
    first_day_values=first_day['Value'].values
    last_day['Value']=last_day['Value']-first_day_values
    last_day=last_day.sort_values(by='Value',ascending=False)
    countries=last_day['Country/Region'].unique()[:3]
    return countries

This function plots the top 3 countries in a subplot

In [23]:
def top_3(week,week_number,subject):
    top_3=top_3_trends(week,week_number)
    week=week.sort_values(by='Date',ascending=True)
    
    subplot_title=[]
    for i in top_3:
        subplot_title.append( '{}'.format(i))
    subplot_title=tuple(subplot_title)
    fig = make_subplots(rows=1, cols=3,subplot_titles=subplot_title)
    
    col=1
    for i in range(0,3):
        temp=week.loc[week['Country/Region']==top_3[i]]
        temp['Value']-=temp.iloc[0,5]
        fig.append_trace(go.Scatter(x=temp['Date'],y=temp['Value'],mode='lines',showlegend=False
                                    ),col=col,row=1)
        
        col+=1
    
    fig.update_layout(title_text="Week {} trends of {} in top 3 countries".format(week_number,subject))
            
    fig.show()
In [24]:
for i in range(1,22):
    week=weekly_trend(confirmed,i)
    top_3(week,i,'Confirmed Cases')
#The same can be done for Recovered and death cases, but the the kernel becomes to long and redundant

C:\Users\thejin\anaconda3\lib\site-packages\ipykernel_launcher.py:5: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\thejin\anaconda3\lib\site-packages\ipykernel_launcher.py:14: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy